Are you curious about how your employment might be affected by artificial intelligence? Look to radiology as a hint.
In the AI race, radiology has recently come into the spotlight. Tech CEOs made repeated references to it last month at the World Economic Forum in Davos and in a White House whitepaper on AI and the economy.
Artificial Intelligence (AI) is having a significant impact on a wide range of professions, including radiology, software engineering, education, and even plumbing. AI developments might replace 6–7% of the US workforce if they are broadly adopted, according to Goldman Sachs, while new employment will also be created by the technology.
However, the profession of radiology is now a case study for how AI could improve rather than replace jobs. Additionally, radiology work is a perfect fit for AI support, according to Dr. Po-Hao Chen, a diagnostic radiology specialist at the Cleveland Clinic.
For AI research and applications, which require large volumes of data for training, radiology offers a wealth of data. Artificial intelligence (AI) is already assisting in the acceleration of some radiology procedures, such as determining which scans require immediate attention, because it can handle vast amounts of data far more quickly than human workers.
However, the majority of the job still needs to be done by human doctors, including diagnosing, physically examining, and writing reports. Due to the field’s continued adoption of technology, radiology jobs are expected to rise faster than jobs in other industries.
Jack Karsten, a research fellow at Georgetown’s Center for Security and Emerging Technology, stated that not only is artificial intelligence (AI) not replacing those professionals, but it is also expanding the amount of work they can accomplish and raising demand for their services. “The tech sector can point to that as a sort of bright future in terms of AI helping the economy.”
How AI complements jobs without taking their place
Radiology relies heavily on AI’s ability to analyze images and identify patterns in data. Additionally, Chen claims that because the industry has been digitalized for years, there is a wealth of data.
“Every X-ray, CT scan, and MRI in the US can be available as zeros and ones, though there are still some smaller use cases that are analogue,” Chen stated.
According to Dr. Chen and two other radiology experts who talked to, radiologists are now utilizing AI to help them prioritize scans, improve image quality, and help them summarize findings.
Interventional radiology specialist Dr. Shadpour Demehri of Johns Hopkins Medicine stated, “It’s something that doesn’t replace anyone, it just makes our job more efficient and meaningful.”
AI is especially helpful for obtaining high-quality MRI images with fewer measurements, according to René Vidal, a professor of engineering and radiology at the University of Pennsylvania’s Penn Engineering department. More people can be seen in the same amount of time thanks to the accelerated process.
Though they’re probably still a ways off, Vidal said, other uses of AI are being investigated in research, such as measuring a tumor’s volume or automatically populating reports.
Jobs that were expected to disappear but didn’t
AI technologies must be approved by the US Food and Drug Administration for medical usage, which may take up to eight years due to the research process and clinical testing, Vidal stated. However, those approvals are happening: 1,041 of the 1,357 AI-enabled medical equipment now approved by the FDA are for radiology.
At the same time, radiology positions appear to be increasing. The Bureau of Labor Statistics predicts that employment in radiology will increase by 5% between 2024 and 2034, outpacing the overall average of 3%. Indeed data released to CNN also reveals that there were more radiology positions in 2025 than five years ago.
According to the radiology experts, the need for more radiology services is probably being driven by the necessity for imaging throughout the medical diagnosis process as well as an aging population.
However, it wasn’t always the case. Geoffrey Hinton, the founder of artificial intelligence and a Nobel Prize-winning computer scientist, stated in 2016 that “people should stop training radiologists now” because deep learning, a branch of AI that simulates how the human brain learns, would be more capable of handling the position in five to ten years.
In an email to the New York Times last year, Hinton claimed that his 2016 remarks were overly general.
According to Demehri, there was concern in the radiology community about artificial intelligence (AI) taking over human jobs in 2015 and 2016. According to him, the device is now viewed as a “second set of eyes.”
The drawbacks of excessive dependence
According to Chen, there are still concerns about bias and possible over-reliance on AI. For instance, according to a 2022 MIT study, AI can correctly identify a person’s race from an X-ray, unlike human radiologists, which raises questions about bias in diagnosis.
Chen is also concerned about the temptation to make staffing decisions, such as replacing a doctor with a nurse or a subspecialist radiologist with a primary care doctor, if AI advances sufficiently. That may be effective in some cases, but not in the vast majority of radiological applications, such as cancer detection or dangerous infections.
We must comprehend that the fact that the automated output is examined by an expert contributes significantly to the algorithm’s performance, he stated. And the collaboration, if you will, between the machine and the expert is what makes the progress possible.
